aboutsummaryrefslogtreecommitdiffstats
path: root/python/fatcat_web/search.py
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@robocracy.org>2018-11-12 23:18:56 -0800
committerBryan Newbold <bnewbold@robocracy.org>2018-11-12 23:18:56 -0800
commitb03bfc8f3fd84141738f775b273a99850d78e1ff (patch)
tree64858e474fa38aa015f06f5e15b851dcc85da421 /python/fatcat_web/search.py
parent055c464deea8cdaccf3ed384995d4409b0f51409 (diff)
downloadfatcat-b03bfc8f3fd84141738f775b273a99850d78e1ff.tar.gz
fatcat-b03bfc8f3fd84141738f775b273a99850d78e1ff.zip
refactor python modules
Diffstat (limited to 'python/fatcat_web/search.py')
-rw-r--r--python/fatcat_web/search.py60
1 files changed, 60 insertions, 0 deletions
diff --git a/python/fatcat_web/search.py b/python/fatcat_web/search.py
new file mode 100644
index 00000000..b6826110
--- /dev/null
+++ b/python/fatcat_web/search.py
@@ -0,0 +1,60 @@
+
+import requests
+from flask import abort
+from fatcat import app
+
+
+def do_search(q, limit=50, fulltext_only=True):
+
+ #print("Search hit: " + q)
+ if limit > 100:
+ # Sanity check
+ limit = 100
+
+ if fulltext_only:
+ q += " file_in_ia:true"
+
+ search_request = {
+ "query": {
+ "query_string": {
+ "query": q,
+ "analyzer": "textIcuSearch",
+ "default_operator": "AND",
+ "analyze_wildcard": True,
+ "lenient": True,
+ "fields": ["title^5", "contrib_names^2", "container_title"]
+ },
+ },
+ "size": int(limit),
+ }
+
+ #print(search_request)
+ resp = requests.get("%s/%s/_search" %
+ (app.config['ELASTIC_BACKEND'], app.config['ELASTIC_INDEX']),
+ json=search_request)
+
+ if resp.status_code != 200:
+ print("elasticsearch non-200 status code: " + str(resp.status_code))
+ print(resp.content)
+ abort(resp.status_code)
+
+ content = resp.json()
+ #print(content)
+ results = [h['_source'] for h in content['hits']['hits']]
+ for h in results:
+ # Ensure 'contrib_names' is a list, not a single string
+ if type(h['contrib_names']) is not list:
+ h['contrib_names'] = [h['contrib_names'], ]
+ # Handle surrogate strings that elasticsearch returns sometimes,
+ # probably due to mangled data processing in some pipeline.
+ # "Crimes against Unicode"; production workaround
+ for key in h:
+ if type(h[key]) is str:
+ h[key] = h[key].encode('utf8', 'ignore').decode('utf8')
+ h['contrib_names'] = [name.encode('utf8', 'ignore').decode('utf8') for name in h['contrib_names']]
+
+ found = content['hits']['total']
+ return {"query": { "q": q },
+ "count_returned": len(results),
+ "count_found": found,
+ "results": results }